Data Flow
A typical request follows this path:
- The user types a question in the frontend
- The frontend sends it to the backend
- The backend forwards it to the external LLM with the available MCP tools in context
- The LLM responds - either with a direct answer, or with one or more tool calls
- Tool calls are routed to the MCP server, which reads from or writes to eMASS
- Results are returned to the LLM to complete its response
- The final answer is passed back through the backend to the frontend